Run 5. Outer Step 143. Inner Step 1.
Browse files- config.json +67 -67
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -12,44 +12,44 @@
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
-
"108": "
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
-
"112": "
|
| 21 |
-
"113": "
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
-
"116": "
|
| 25 |
-
"117": "
|
| 26 |
-
"118": "
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
-
"121": "
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
-
"124": "
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
-
"127": "
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
-
"130": "
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
-
"136": "
|
| 47 |
-
"137": "
|
| 48 |
-
"138": "
|
| 49 |
-
"139": "
|
| 50 |
-
"14": "
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
-
"141": "
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
|
@@ -64,8 +64,8 @@
|
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
-
"155": "
|
| 68 |
-
"156": "
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
|
@@ -76,8 +76,8 @@
|
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
-
"166": "
|
| 80 |
-
"167": "
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
@@ -85,7 +85,7 @@
|
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
-
"174": "
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
@@ -95,51 +95,51 @@
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
-
"183": "
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
-
"187": "
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
-
"193": "
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
-
"195": "
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
-
"197": "
|
| 114 |
-
"198": "
|
| 115 |
-
"199": "
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
-
"202": "
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
-
"204": "
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
-
"206": "
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
-
"212": "
|
| 132 |
-
"213": "
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
-
"215": "
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
-
"217": "
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
-
"221": "
|
| 142 |
-
"222": "
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NON_PARTICIPATING",
|
|
@@ -147,43 +147,43 @@
|
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
-
"23": "
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
-
"231": "
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
-
"236": "
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
-
"238": "
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
-
"241": "
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
-
"243": "
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
-
"245": "
|
| 168 |
"246": "NOT_ALIVE",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
-
"250": "
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
-
"254": "
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
-
"30": "
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
-
"32": "
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
@@ -191,20 +191,20 @@
|
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
-
"4": "
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
-
"44": "
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
-
"46": "
|
| 202 |
-
"47": "
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
-
"5": "
|
| 206 |
-
"50": "
|
| 207 |
-
"51": "
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
|
@@ -212,7 +212,7 @@
|
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
-
"59": "
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
|
@@ -228,30 +228,30 @@
|
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
-
"73": "
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
-
"77": "
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
-
"8": "
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
-
"85": "
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
-
"91": "
|
| 252 |
-
"92": "
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
-
"94": "
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "NON_PARTICIPATING",
|
|
@@ -264,17 +264,17 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
-
"inner_step":
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
-
"last_allreduce_block":
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
|
|
|
| 12 |
"105": "NON_PARTICIPATING",
|
| 13 |
"106": "NON_PARTICIPATING",
|
| 14 |
"107": "NOT_ALIVE",
|
| 15 |
+
"108": "NON_PARTICIPATING",
|
| 16 |
"109": "NON_PARTICIPATING",
|
| 17 |
"11": "NOT_ALIVE",
|
| 18 |
"110": "NOT_ALIVE",
|
| 19 |
"111": "NON_PARTICIPATING",
|
| 20 |
+
"112": "NON_PARTICIPATING",
|
| 21 |
+
"113": "FAIL",
|
| 22 |
"114": "NOT_ALIVE",
|
| 23 |
"115": "NON_PARTICIPATING",
|
| 24 |
+
"116": "NON_PARTICIPATING",
|
| 25 |
+
"117": "NON_PARTICIPATING",
|
| 26 |
+
"118": "NON_PARTICIPATING",
|
| 27 |
"119": "NON_PARTICIPATING",
|
| 28 |
"12": "NOT_ALIVE",
|
| 29 |
"120": "NON_PARTICIPATING",
|
| 30 |
+
"121": "NOT_ALIVE",
|
| 31 |
"122": "NON_PARTICIPATING",
|
| 32 |
"123": "NON_PARTICIPATING",
|
| 33 |
+
"124": "NOT_ALIVE",
|
| 34 |
"125": "NON_PARTICIPATING",
|
| 35 |
"126": "NOT_ALIVE",
|
| 36 |
+
"127": "NON_PARTICIPATING",
|
| 37 |
"128": "NON_PARTICIPATING",
|
| 38 |
"129": "NON_PARTICIPATING",
|
| 39 |
"13": "NOT_ALIVE",
|
| 40 |
+
"130": "NOT_ALIVE",
|
| 41 |
"131": "NOT_ALIVE",
|
| 42 |
"132": "NON_PARTICIPATING",
|
| 43 |
"133": "NON_PARTICIPATING",
|
| 44 |
"134": "NON_PARTICIPATING",
|
| 45 |
"135": "NON_PARTICIPATING",
|
| 46 |
+
"136": "NOT_ALIVE",
|
| 47 |
+
"137": "NON_PARTICIPATING",
|
| 48 |
+
"138": "SUCCESS",
|
| 49 |
+
"139": "NON_PARTICIPATING",
|
| 50 |
+
"14": "NON_PARTICIPATING",
|
| 51 |
"140": "NON_PARTICIPATING",
|
| 52 |
+
"141": "NON_PARTICIPATING",
|
| 53 |
"142": "NOT_ALIVE",
|
| 54 |
"143": "NON_PARTICIPATING",
|
| 55 |
"144": "NOT_ALIVE",
|
|
|
|
| 64 |
"152": "NON_PARTICIPATING",
|
| 65 |
"153": "NOT_ALIVE",
|
| 66 |
"154": "NON_PARTICIPATING",
|
| 67 |
+
"155": "FAIL",
|
| 68 |
+
"156": "NON_PARTICIPATING",
|
| 69 |
"157": "NOT_ALIVE",
|
| 70 |
"158": "NOT_ALIVE",
|
| 71 |
"159": "NOT_ALIVE",
|
|
|
|
| 76 |
"163": "NOT_ALIVE",
|
| 77 |
"164": "NOT_ALIVE",
|
| 78 |
"165": "NOT_ALIVE",
|
| 79 |
+
"166": "NOT_ALIVE",
|
| 80 |
+
"167": "NON_PARTICIPATING",
|
| 81 |
"168": "NON_PARTICIPATING",
|
| 82 |
"169": "NOT_ALIVE",
|
| 83 |
"17": "NOT_ALIVE",
|
|
|
|
| 85 |
"171": "NOT_ALIVE",
|
| 86 |
"172": "NON_PARTICIPATING",
|
| 87 |
"173": "NON_PARTICIPATING",
|
| 88 |
+
"174": "NOT_ALIVE",
|
| 89 |
"175": "NOT_ALIVE",
|
| 90 |
"176": "NOT_ALIVE",
|
| 91 |
"177": "NOT_ALIVE",
|
|
|
|
| 95 |
"180": "NOT_ALIVE",
|
| 96 |
"181": "NOT_ALIVE",
|
| 97 |
"182": "NOT_ALIVE",
|
| 98 |
+
"183": "NON_PARTICIPATING",
|
| 99 |
"184": "NON_PARTICIPATING",
|
| 100 |
"185": "NOT_ALIVE",
|
| 101 |
"186": "NOT_ALIVE",
|
| 102 |
+
"187": "NOT_ALIVE",
|
| 103 |
"188": "NON_PARTICIPATING",
|
| 104 |
"189": "NOT_ALIVE",
|
| 105 |
"19": "NON_PARTICIPATING",
|
| 106 |
"190": "NON_PARTICIPATING",
|
| 107 |
"191": "NON_PARTICIPATING",
|
| 108 |
"192": "NON_PARTICIPATING",
|
| 109 |
+
"193": "NON_PARTICIPATING",
|
| 110 |
"194": "NON_PARTICIPATING",
|
| 111 |
+
"195": "NON_PARTICIPATING",
|
| 112 |
"196": "NON_PARTICIPATING",
|
| 113 |
+
"197": "NON_PARTICIPATING",
|
| 114 |
+
"198": "SUCCESS",
|
| 115 |
+
"199": "NOT_ALIVE",
|
| 116 |
"2": "NON_PARTICIPATING",
|
| 117 |
"20": "NOT_ALIVE",
|
| 118 |
"200": "NOT_ALIVE",
|
| 119 |
"201": "NOT_ALIVE",
|
| 120 |
+
"202": "NOT_ALIVE",
|
| 121 |
"203": "NOT_ALIVE",
|
| 122 |
+
"204": "SUCCESS",
|
| 123 |
"205": "NON_PARTICIPATING",
|
| 124 |
+
"206": "SUCCESS",
|
| 125 |
"207": "NOT_ALIVE",
|
| 126 |
"208": "NON_PARTICIPATING",
|
| 127 |
"209": "NOT_ALIVE",
|
| 128 |
"21": "NOT_ALIVE",
|
| 129 |
"210": "NON_PARTICIPATING",
|
| 130 |
"211": "NON_PARTICIPATING",
|
| 131 |
+
"212": "NON_PARTICIPATING",
|
| 132 |
+
"213": "NOT_ALIVE",
|
| 133 |
"214": "NON_PARTICIPATING",
|
| 134 |
+
"215": "NON_PARTICIPATING",
|
| 135 |
"216": "NOT_ALIVE",
|
| 136 |
+
"217": "NON_PARTICIPATING",
|
| 137 |
"218": "NOT_ALIVE",
|
| 138 |
"219": "NOT_ALIVE",
|
| 139 |
"22": "NON_PARTICIPATING",
|
| 140 |
"220": "NON_PARTICIPATING",
|
| 141 |
+
"221": "NON_PARTICIPATING",
|
| 142 |
+
"222": "NON_PARTICIPATING",
|
| 143 |
"223": "NOT_ALIVE",
|
| 144 |
"224": "NOT_ALIVE",
|
| 145 |
"225": "NON_PARTICIPATING",
|
|
|
|
| 147 |
"227": "NOT_ALIVE",
|
| 148 |
"228": "NOT_ALIVE",
|
| 149 |
"229": "NON_PARTICIPATING",
|
| 150 |
+
"23": "NOT_ALIVE",
|
| 151 |
"230": "NON_PARTICIPATING",
|
| 152 |
+
"231": "NOT_ALIVE",
|
| 153 |
"232": "NOT_ALIVE",
|
| 154 |
"233": "NOT_ALIVE",
|
| 155 |
"234": "NOT_ALIVE",
|
| 156 |
"235": "NON_PARTICIPATING",
|
| 157 |
+
"236": "NON_PARTICIPATING",
|
| 158 |
"237": "NON_PARTICIPATING",
|
| 159 |
+
"238": "NOT_ALIVE",
|
| 160 |
"239": "NOT_ALIVE",
|
| 161 |
"24": "NOT_ALIVE",
|
| 162 |
"240": "NON_PARTICIPATING",
|
| 163 |
+
"241": "NON_PARTICIPATING",
|
| 164 |
"242": "NON_PARTICIPATING",
|
| 165 |
+
"243": "NOT_ALIVE",
|
| 166 |
"244": "NOT_ALIVE",
|
| 167 |
+
"245": "NOT_ALIVE",
|
| 168 |
"246": "NOT_ALIVE",
|
| 169 |
"247": "NON_PARTICIPATING",
|
| 170 |
"248": "NON_PARTICIPATING",
|
| 171 |
"249": "NON_PARTICIPATING",
|
| 172 |
"25": "SUCCESS",
|
| 173 |
+
"250": "FAIL",
|
| 174 |
"251": "NOT_ALIVE",
|
| 175 |
"252": "NOT_ALIVE",
|
| 176 |
"253": "NOT_ALIVE",
|
| 177 |
+
"254": "NOT_ALIVE",
|
| 178 |
"255": "NOT_ALIVE",
|
| 179 |
"26": "NON_PARTICIPATING",
|
| 180 |
"27": "NOT_ALIVE",
|
| 181 |
"28": "NOT_ALIVE",
|
| 182 |
"29": "NON_PARTICIPATING",
|
| 183 |
"3": "NON_PARTICIPATING",
|
| 184 |
+
"30": "SUCCESS",
|
| 185 |
"31": "NON_PARTICIPATING",
|
| 186 |
+
"32": "NOT_ALIVE",
|
| 187 |
"33": "NOT_ALIVE",
|
| 188 |
"34": "NOT_ALIVE",
|
| 189 |
"35": "NON_PARTICIPATING",
|
|
|
|
| 191 |
"37": "NOT_ALIVE",
|
| 192 |
"38": "NON_PARTICIPATING",
|
| 193 |
"39": "SUCCESS",
|
| 194 |
+
"4": "NON_PARTICIPATING",
|
| 195 |
"40": "NON_PARTICIPATING",
|
| 196 |
"41": "NON_PARTICIPATING",
|
| 197 |
"42": "NOT_ALIVE",
|
| 198 |
"43": "NOT_ALIVE",
|
| 199 |
+
"44": "NON_PARTICIPATING",
|
| 200 |
"45": "NOT_ALIVE",
|
| 201 |
+
"46": "NON_PARTICIPATING",
|
| 202 |
+
"47": "NON_PARTICIPATING",
|
| 203 |
"48": "NON_PARTICIPATING",
|
| 204 |
"49": "NON_PARTICIPATING",
|
| 205 |
+
"5": "SUCCESS",
|
| 206 |
+
"50": "NON_PARTICIPATING",
|
| 207 |
+
"51": "NON_PARTICIPATING",
|
| 208 |
"52": "SUCCESS",
|
| 209 |
"53": "NON_PARTICIPATING",
|
| 210 |
"54": "NOT_ALIVE",
|
|
|
|
| 212 |
"56": "NON_PARTICIPATING",
|
| 213 |
"57": "NOT_ALIVE",
|
| 214 |
"58": "NON_PARTICIPATING",
|
| 215 |
+
"59": "NOT_ALIVE",
|
| 216 |
"6": "NON_PARTICIPATING",
|
| 217 |
"60": "NOT_ALIVE",
|
| 218 |
"61": "NON_PARTICIPATING",
|
|
|
|
| 228 |
"70": "NOT_ALIVE",
|
| 229 |
"71": "NON_PARTICIPATING",
|
| 230 |
"72": "NON_PARTICIPATING",
|
| 231 |
+
"73": "NON_PARTICIPATING",
|
| 232 |
"74": "NOT_ALIVE",
|
| 233 |
"75": "NON_PARTICIPATING",
|
| 234 |
"76": "NOT_ALIVE",
|
| 235 |
+
"77": "NON_PARTICIPATING",
|
| 236 |
"78": "NON_PARTICIPATING",
|
| 237 |
"79": "NON_PARTICIPATING",
|
| 238 |
+
"8": "NON_PARTICIPATING",
|
| 239 |
"80": "NOT_ALIVE",
|
| 240 |
"81": "NON_PARTICIPATING",
|
| 241 |
"82": "NON_PARTICIPATING",
|
| 242 |
"83": "NON_PARTICIPATING",
|
| 243 |
"84": "NON_PARTICIPATING",
|
| 244 |
+
"85": "NOT_ALIVE",
|
| 245 |
"86": "NON_PARTICIPATING",
|
| 246 |
"87": "NOT_ALIVE",
|
| 247 |
"88": "NON_PARTICIPATING",
|
| 248 |
"89": "NOT_ALIVE",
|
| 249 |
"9": "NON_PARTICIPATING",
|
| 250 |
"90": "NOT_ALIVE",
|
| 251 |
+
"91": "NON_PARTICIPATING",
|
| 252 |
+
"92": "NON_PARTICIPATING",
|
| 253 |
"93": "NOT_ALIVE",
|
| 254 |
+
"94": "FAIL",
|
| 255 |
"95": "NOT_ALIVE",
|
| 256 |
"96": "NOT_ALIVE",
|
| 257 |
"97": "NON_PARTICIPATING",
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6019232,
|
| 268 |
+
6019259
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
+
"inner_step": 1,
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
+
"last_allreduce_block": 6019128,
|
| 278 |
"max_position_embeddings": 2048,
|
| 279 |
"mlp_bias": false,
|
| 280 |
"model_type": "llama",
|
gradients.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 44585978
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:461b9363856469ee7654e332651b67918d755b864d235f76dcea6a11aa753b97
|
| 3 |
size 44585978
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8800564076
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bb019d69e09c04140886017762bcfa73bf51466d7b17efa37db478ca4c35fab3
|
| 3 |
size 8800564076
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4400216536
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:88c294a4b1cccd06d859a5fbdf777fba96747f4e6d792b2023339b68bbe4a5d0
|
| 3 |
size 4400216536
|